Libre RISC-V Mailing list Archives (accessible via public-inbox)
 help / color / Atom feed
* Re: [libre-riscv-dev] daily kan-ban update 23may2020
From: Cesar Strauss @ 2020-05-23 22:08 UTC (permalink / raw)
  To: libre-riscv-dev
In-Reply-To: <CAPweEDzcvyqr-25ctDUB38zkDtjXMWX9xErAh8U77QrdyuyXzA@mail.gmail.com>

On 05/23/2020 17:57, Luke Kenneth Casson Leighton escreveu:
> it is 3am in brazil so Cesar will not be up, write on that bugreport when
> you start and finish ok?

Something must have gone wrong somewhere...

It was only around 18:00 here, when you wrote that, not 03:00.

I'm on UTC-03:00.

Try:

$ TZ='America/Sao_Paulo' date

or, do "date -u" and subtract 3 hours.

Cheers,

Cesar

_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev

^ permalink raw reply

* Re: [libre-riscv-dev] daily kan-ban update 23may2020
From: Cole Poirier @ 2020-05-23 21:38 UTC (permalink / raw)
  To: Libre-RISCV General Development; +Cc: Libre-RISCV General Development
In-Reply-To: <CAPweEDw1nSkxL084QXyzNHGf+9hmi6ZTkgaV1F4xAkW4EJi2ug@mail.gmail.com>

On May 23 2020, at 1:59 pm, Luke Kenneth Casson Leighton <lkcl@lkcl.net> wrote:
> soc.fu they are now common base classes.  the code is exactly the
> same, it
> is just split.

I tried fixing this the naive way, by just substituting CommonInput
stageStage for LogicalInputStage but this results in the following error:

```
File "test_pipe_caller.py", line 193, in run_all
  m.submodules.alu = alu = LogicalBasePipe(pspec)
File "/home/colepoirier/src/soc/src/soc/fu/logical/pipeline.py", line
18, in __init__
  self.pipe1 = LogicalStages(pspec)
File "/home/colepoirier/src/nmutil/src/nmutil/dynamicpipe.py", line 37,
in __call__
 return super().__call__(*args, **kw)
File "/home/colepoirier/src/nmutil/src/nmutil/pipemodbase.py", line 34,
in __init__
  self.chain = self.get_chain()
File "/home/colepoirier/src/soc/src/soc/fu/logical/pipeline.py", line 9,
in get_chain
  inp = CommonInputStage(self.pspec)
TypeError: __init__() missing 1 required positional argument: 'modname'
```

Do I have to add something from fu/logical/logical_input_record.py, or
add something else in order to use CommonInputStage properly?


> one weirdness, if you put a print statement on a and b, a is a signal, but
> b ends up as an actual integer.

Are you referring to a and b of the CommonStage's or of another piece of code?

Cole


_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev

^ permalink raw reply

* Re: [libre-riscv-dev] daily kan-ban update 23may2020
From: Cole Poirier @ 2020-05-23 21:21 UTC (permalink / raw)
  To: Libre-RISCV General Development; +Cc: Libre-RISCV General Development
In-Reply-To: <CAPweEDzcvyqr-25ctDUB38zkDtjXMWX9xErAh8U77QrdyuyXzA@mail.gmail.com>

On May 23 2020, at 1:36 pm, Michael Nolan <mtnolan2640@gmail.com> wrote:
> I took a look at it earlier, but the test was broken because the input 
> and output stages got deleted/moved, and I can't figure out what needs 
> to be used to replace them.
> 
> --Michael.

On May 23 2020, at 1:57 pm, Luke Kenneth Casson Leighton <lkcl@lkcl.net> wrote:
> actually you could look at adding zero_a to op_sim in compalu_multi.  see
> https://bugs.libre-soc.org/show_bug.cgi?id=336
> 
> it is 3am in brazil so Cesar will not be up, write on that bugreport when
> you start and finish ok?
> 
> so in the morning he can take over again.
> 
> l.

Thanks Luke and Michael. My plan for today then is to fix the tests for
the logical pipeline, then notify Michael of this task's completion on
the BPERM TODO bug report so he can take a look at what I've done wrong
in bpermd. After fixing the fu/logical/tests/test_pipe_caller.py, I will
work on adding zero_a to op_sim in compalu_multi on bug 336.

Cole

_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev

^ permalink raw reply

* [libre-riscv-dev] [Bug 314] Create POWER9 Condition Register pipeline
From: bugzilla-daemon @ 2020-05-23 21:01 UTC (permalink / raw)
  To: libre-riscv-dev
In-Reply-To: <bug-314-13@https.bugs.libre-soc.org/>

https://bugs.libre-soc.org/show_bug.cgi?id=314

--- Comment #20 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
hm might be good to add the test assembly code to the repo, put it somewhere
around soc.fu.cr

-- 
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev

^ permalink raw reply

* Re: [libre-riscv-dev] daily kan-ban update 23may2020
From: Luke Kenneth Casson Leighton @ 2020-05-23 20:59 UTC (permalink / raw)
  To: Libre-RISCV General Development
In-Reply-To: <0fdd33fe-9fd5-2f55-2524-cbf8ba24d8e3@gmail.com>

On Saturday, May 23, 2020, Michael Nolan <mtnolan2640@gmail.com> wrote:

>
> I took a look at it earlier, but the test was broken because the input and
> output stages got deleted/moved, and I can't figure out what needs to be
> used to replace them.


soc.fu they are now common base classes.  the code is exactly the same, it
is just split.

one weirdness, if you put a print statement on a and b, a is a signal, but
b ends up as an actual integer.

l.



-- 
---
crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev

^ permalink raw reply

* Re: [libre-riscv-dev] daily kan-ban update 23may2020
From: Luke Kenneth Casson Leighton @ 2020-05-23 20:57 UTC (permalink / raw)
  To: Libre-RISCV General Development
In-Reply-To: <152B0355-0711-4C08-B509-7C1BEEC61357@gmail.com>

On Saturday, May 23, 2020, Cole Poirier <colepoirier@gmail.com> wrote:

> Luke, since I am blocked on bpermd until Michael is able to take a look,
> and he is currently busy with the CR pipeline proofs, is there a new
> instruction I can work on implementing? Or tests that need to be written
> for existing modules that you’d like me to try doing?


actually you could look at adding zero_a to op_sim in compalu_multi.  see
https://bugs.libre-soc.org/show_bug.cgi?id=336

it is 3am in brazil so Cesar will not be up, write on that bugreport when
you start and finish ok?

so in the morning he can take over again.

l.



-- 
---
crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev

^ permalink raw reply

* [libre-riscv-dev] [Bug 336] ALU CompUnit needs to recognise that RA (src1) can be zero
From: bugzilla-daemon @ 2020-05-23 20:52 UTC (permalink / raw)
  To: libre-riscv-dev
In-Reply-To: <bug-336-13@https.bugs.libre-soc.org/>

https://bugs.libre-soc.org/show_bug.cgi?id=336

--- Comment #10 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
cesar i created a common function for both zero a and imm b muxing.  should be
functionally the same.

note that the Mux will automatically have its inputs extended with zero padding
to the max operand width.

so i replaced Const 0 with just 0

-- 
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev

^ permalink raw reply

* Re: [libre-riscv-dev] daily kan-ban update 23may2020
From: Michael Nolan @ 2020-05-23 20:36 UTC (permalink / raw)
  To: libre-riscv-dev
In-Reply-To: <152B0355-0711-4C08-B509-7C1BEEC61357@gmail.com>

On 5/23/20 4:15 PM, Cole Poirier wrote:
> Luke, since I am blocked on bpermd until Michael is able to take a look, and he is currently busy with the CR pipeline proofs, is there a new instruction I can work on implementing? Or tests that need to be written for existing modules that you’d like me to try doing? Essentially what tasks still need to be completed for the Oct 2020 prototype outside of the very technical and complex pieces like those that you, Michael, Jacob, Cesar, et al are focussing on? I realize that this information should be discernible from the wiki, bug reports and mailing list, however, due to my inexperience I’d like to ask for you guidance on what would be most helpful in achieving our goals for October.
>
> Cole
>
>
>
> _______________________________________________
> libre-riscv-dev mailing list
> libre-riscv-dev@lists.libre-riscv.org
> http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev

I took a look at it earlier, but the test was broken because the input 
and output stages got deleted/moved, and I can't figure out what needs 
to be used to replace them.

--Michael.


_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev

^ permalink raw reply

* [libre-riscv-dev] [Bug 314] Create POWER9 Condition Register pipeline
From: bugzilla-daemon @ 2020-05-23 20:30 UTC (permalink / raw)
  To: libre-riscv-dev
In-Reply-To: <bug-314-13@https.bugs.libre-soc.org/>

https://bugs.libre-soc.org/show_bug.cgi?id=314

--- Comment #19 from Michael Nolan <mtnolan2640@gmail.com> ---
Created attachment 62
  --> https://bugs.libre-soc.org/attachment.cgi?id=62&action=edit
table of mtocrf mask, reg results from the talos server

So I decided to test mtocrf with non-onehot masks on the talos server, and it
appears to me that it uses the most significant bit of the mask and ignores the
rest (except for mask=0xf, dunno what's up there). Very interesting.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev

^ permalink raw reply

* Re: [libre-riscv-dev] daily kan-ban update 23may2020
From: Cole Poirier @ 2020-05-23 20:15 UTC (permalink / raw)
  To: Libre-RISCV General Development
In-Reply-To: <CAPweEDyi5Dep=o3STGRg_bVc8w2jJEiVF7NJyyp3WjW7dgr54Q@mail.gmail.com>

Luke, since I am blocked on bpermd until Michael is able to take a look, and he is currently busy with the CR pipeline proofs, is there a new instruction I can work on implementing? Or tests that need to be written for existing modules that you’d like me to try doing? Essentially what tasks still need to be completed for the Oct 2020 prototype outside of the very technical and complex pieces like those that you, Michael, Jacob, Cesar, et al are focussing on? I realize that this information should be discernible from the wiki, bug reports and mailing list, however, due to my inexperience I’d like to ask for you guidance on what would be most helpful in achieving our goals for October.

Cole



_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev

^ permalink raw reply

* [libre-riscv-dev] [Bug 336] ALU CompUnit needs to recognise that RA (src1) can be zero
From: bugzilla-daemon @ 2020-05-23 19:39 UTC (permalink / raw)
  To: libre-riscv-dev
In-Reply-To: <bug-336-13@https.bugs.libre-soc.org/>

https://bugs.libre-soc.org/show_bug.cgi?id=336

--- Comment #9 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
(In reply to Luke Kenneth Casson Leighton from comment #8)
> (In reply to Cesar Strauss from comment #7)
> > (In reply to Luke Kenneth Casson Leighton from comment #6)
> > > ok done for imm_data, here is the location, it is virtually a cookie-cut of
> > > what
> > > is done for imm_data, bear in mind it will need to be src_l.q[0] and because
> > > src_l.q[0] covers RA, where src_l.q[1] covers RB.  likewise the src1_or_imm
> > > needs to go into sl[0] not sl[1]
> > 
> > I have done the above.
> 
> excellent.
> 
> > I held my commit because you seemed to be working on the file.
> 
> yeh don't do that: just shove it in :)
> 
> > It's OK to commit now?
> 
> git pull to merge, first, then go for it.  i happen to just be going out
> shopping
> anyway.

looks great, cesar - unit test time.  it should be as simple as setting
dut.oper_i.zero_a.eq(1) - or adding an extra argument to op_sim() to
allow that to be set dynamically - then setting a to a non-zero value
and checking that, yes, actually, a is now irrelevant (set to zero).

a more advanced version of the op_sim() test should be not to set
rd.go.eq(0b11)
in an ad-hoc fashion: instead to check rd_rel_o bits.

this starts to be very much like the planned LD/ST unit test modifications,
and you could probably use the exact same code for both.

however please do everything in an incremental fashion: no more than 5-10/15
lines at a time when modifying existing code (if that is practical).

so, start with adding the extra argument to op_sim() to allow zero_a to be set

then add a test call in scoreboard_sim op_sim(a=12312313213, zero_a=True, ...)

and so on.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev

^ permalink raw reply

* [libre-riscv-dev] [Bug 336] ALU CompUnit needs to recognise that RA (src1) can be zero
From: bugzilla-daemon @ 2020-05-23 19:16 UTC (permalink / raw)
  To: libre-riscv-dev
In-Reply-To: <bug-336-13@https.bugs.libre-soc.org/>

https://bugs.libre-soc.org/show_bug.cgi?id=336

--- Comment #8 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
(In reply to Cesar Strauss from comment #7)
> (In reply to Luke Kenneth Casson Leighton from comment #6)
> > ok done for imm_data, here is the location, it is virtually a cookie-cut of
> > what
> > is done for imm_data, bear in mind it will need to be src_l.q[0] and because
> > src_l.q[0] covers RA, where src_l.q[1] covers RB.  likewise the src1_or_imm
> > needs to go into sl[0] not sl[1]
> 
> I have done the above.

excellent.

> I held my commit because you seemed to be working on the file.

yeh don't do that: just shove it in :)

> It's OK to commit now?

git pull to merge, first, then go for it.  i happen to just be going out
shopping
anyway.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev

^ permalink raw reply

* [libre-riscv-dev] [Bug 336] ALU CompUnit needs to recognise that RA (src1) can be zero
From: bugzilla-daemon @ 2020-05-23 19:09 UTC (permalink / raw)
  To: libre-riscv-dev
In-Reply-To: <bug-336-13@https.bugs.libre-soc.org/>

https://bugs.libre-soc.org/show_bug.cgi?id=336

--- Comment #7 from Cesar Strauss <cestrauss@gmail.com> ---
(In reply to Luke Kenneth Casson Leighton from comment #6)
> ok done for imm_data, here is the location, it is virtually a cookie-cut of
> what
> is done for imm_data, bear in mind it will need to be src_l.q[0] and because
> src_l.q[0] covers RA, where src_l.q[1] covers RB.  likewise the src1_or_imm
> needs to go into sl[0] not sl[1]

I have done the above. I held my commit because you seemed to be working on the
file. It's OK to commit now?

-- 
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev

^ permalink raw reply

* [libre-riscv-dev] [Bug 332] Formal correctness proof needed for CR pipeline
From: bugzilla-daemon @ 2020-05-23 18:43 UTC (permalink / raw)
  To: libre-riscv-dev
In-Reply-To: <bug-332-13@https.bugs.libre-soc.org/>

https://bugs.libre-soc.org/show_bug.cgi?id=332

--- Comment #7 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
(In reply to Luke Kenneth Casson Leighton from comment #6)

> ok, cool.  we do now have two CR_ISEL case statements,

ok i see, the first one is for setup, the second for actual isel "proof".

i will delete one of the test_isel()s.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev

^ permalink raw reply

* [libre-riscv-dev] Function Units "patch-up" linking to Comp Unit code
From: Luke Kenneth Casson Leighton @ 2020-05-23 18:23 UTC (permalink / raw)
  To: Libre-RISCV General Development

right.  finally.  with regspecs in place, and the pipelines in place,
*finally* we can link them up together:
https://git.libre-soc.org/?p=soc.git;a=blob;f=src/soc/fu/compunits.py;hb=HEAD

MultiCompUnit has been "morphed" to be completely FSM / Pipeline
agnostic.  it knows *nothing* about the data it's dealing with: names
or quantity of input signals, names or quantity of output signals,
absolutely nothing.  (oh, Cesar, Michael: except for the fact that RA
can be zero, and RB can be an immediate).

the *only* way to tell MultiCompUnit the information about the ALU
it's been asked to look after: regspecs.  i've documented the format,
here:
https://libre-soc.org/3d_gpu/architecture/regfile/

any questions please ask.

l.

_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev

^ permalink raw reply

* Re: [libre-riscv-dev] daily kan-ban update 22may2020
From: Cole Poirier @ 2020-05-23 18:16 UTC (permalink / raw)
  To: Libre-RISCV General Development
In-Reply-To: <a1e0dff9-477f-59a0-d5da-6b0b97d75e16@gmail.com>

> On May 22, 2020, at 6:56 PM, Cesar Strauss <cestrauss@gmail.com> wrote:
> 
> Hi Cole and all,
> 
>> On 05/22/2020 16:10, Cole Poirier wrote:
>> Would you mind giving a one line description here of what Cesar does?
>> I'd be interested to know as well.
> 
> I work in a research institute, on data acquisition and control of
> scientific instruments for astronomy.
> 
> Let me expand a bit on that.
> 
> Some of the projects I'm involved, past and present:
> 
> 1) Microwave receivers for a 14 m radio telescope.
> 2) Control of a 26-element radio interferometer array
> 3) Image acquisition for an infrared camera
> 4) Data acquisition for a X-ray detector in a stratospheric balloon.
> 5) Miniaturized version of the above, to fly on a nano-satellite (2U
> cubesat)
> 6) Spherical-mass gravitational wave detector
> 
> Some phenomena that the scientists observe with the above:
> 
> 1) Molecular clouds and star formation
> 2) Variability in the active nuclei of galaxies (quasars)
> 3) Accretion disks around black holes
> 4) Gamma ray burst
> 5) Solar flares and coronal mass ejections
> 6) Space weather
> 
> Some technologies I use:
> 
> 1) Data acquisition and remote control of voltmeters and spectrum
> analyzers via RS-232, GPIB, Ethernet and USB
> 2) Multi-channel analog data acquisition boards
> 3) Microcontrollers like 8051, PIC, ARM Cortex, Arduinos
> 4) FPGA for glue logic, data acquisition and signal processing
> 5) Single-board computer
> 6) Cryogenics for low noise temperature radio frequency receivers
> 
> Software tools:
> 
> 1) GCC, C/C++ compiler for the PC and some microcontrollers
> 2) SDCC, C compiler for smaller microcontrollers
> 3) Icarus Verilog for FPGA simulation
> 4) gEDA/gaf, Pcb and Gerbv for PCB design
> 5) Python, for quick prototyping, data processing, plotting.
> 6) R, Octave, for data processing
> 
> My motivation to participate in Libre-SOC:
> 
> A libre, supported, fully documented SoC, free of proprietary drivers,
> firmware and bootloaders, that I can trust to use in my workstation,
> server and as building blocks of scientific instruments.
> 
> Regards,
> 
> Cesar

That’s some wildly cool stuff Cesar! Thanks for sharing, I’m excited we’ve got you as a new member!
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev

^ permalink raw reply

* [libre-riscv-dev] [Bug 216] LOAD STORE buffer needed
From: bugzilla-daemon @ 2020-05-23 15:23 UTC (permalink / raw)
  To: libre-riscv-dev
In-Reply-To: <bug-216-13@https.bugs.libre-soc.org/>

https://bugs.libre-soc.org/show_bug.cgi?id=216

--- Comment #40 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
Tobias: i've added a stub class "DataMerger".  please can you focus on
implementing
that.  it should be approximately... 30 lines of code, maximum.  start by
committing
a class with the Record Layout to be used by the input (Nx) and output (1x)

-- 
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev

^ permalink raw reply

* [libre-riscv-dev] [Bug 216] LOAD STORE buffer needed
From: bugzilla-daemon @ 2020-05-23 15:17 UTC (permalink / raw)
  To: libre-riscv-dev
In-Reply-To: <bug-216-13@https.bugs.libre-soc.org/>

https://bugs.libre-soc.org/show_bug.cgi?id=216

--- Comment #39 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
see 
http://lists.libre-riscv.org/pipermail/libre-riscv-dev/2020-May/007196.html


a new class in soc/experiment/l0_cache.py is needed which
implements that core algorithm: the one that uses a PriorityPicker to
select one of the "active" LD/STs, then use that to OR-select an
output.

please assume the following inputs:

* an array of 8 Signals of length 8 bits which tell you which
addresses have matched against which other addresses.
  ( we call this the "address-match" array)
* an array of 8 Records containing {16 bytes data, 16 "enable" lines}

and the following output:

* a Record containing {16 bytes data, 16 "enable" lines}

in the address-match array, you want to find the first non-zero line.
once you have that index, use it to OR the 16 bytes and 16 "enables"
from the 8 records, into the output.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev

^ permalink raw reply

* Re: [libre-riscv-dev] daily kan-ban update 23may2020
From: Luke Kenneth Casson Leighton @ 2020-05-23 15:15 UTC (permalink / raw)
  To: Libre-RISCV General Development
In-Reply-To: <20200523165620.92b34df777af2e7b1c7812ab@platen-software.de>

On Sat, May 23, 2020 at 3:56 PM Tobias Platen
<libre-soc@platen-software.de> wrote:
>
> If no one else is working on the LDST unit I will start working there.

good.  please however listen, pay attention, keep in constant
communication (do not wait until the end of the day if you are stuck,
like yesterday).

> First I'll copy the skeleton files from another function unit,

that will not be necessary: LDST is special-cased.  LDST is not
pipelined (they are FSMs), so virtually none of the infrastructure
there in soc.fu/* applies to it.

the reason why i created soc.fu.ldst.ldst_input_record is to make it
"consistent" when it comes to wiring up the Computation Units, which
will be dropped into soc/fu/{somefile.py} later.

> then I'll
> try to implement the things described in the video L0 cache/buffer walkthrough [1].

there are several pieces to this.  i need you to focus on them one at
a time, as this is particularly tricky.

first, a new class in soc/experiment/l0_cache.py is needed which
implements that core algorithm: the one that uses a PriorityPicker to
select one of the "active" LD/STs, then use that to OR-select an
output.

please assume the following inputs:

* an array of 8 Signals of length 8 bits which tell you which
addresses have matched against which other addresses.
  ( we call this the "address-match" array)
* an array of 8 Records containing {16 bytes data, 16 "enable" lines}

and the following output:

* a Record containing {16 bytes data, 16 "enable" lines}

in the address-match array, you want to find the first non-zero line.
once you have that index, use it to OR the 16 bytes and 16 "enables"
from the 8 records, into the output.

that's it.  it's quite simple, and is a crucial piece that we need.

i will cut/paste this into the appropriate bugreport (once i have found it).

please do *NOT* modify, replace or remove the existing L0CacheBuffer.
we need it, because it is currently in use.  it is not
high-performance, however it "works", and the rest of the code relies
on it.

l.

_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev

^ permalink raw reply

* Re: [libre-riscv-dev] daily kan-ban update 23may2020
From: Tobias Platen @ 2020-05-23 14:56 UTC (permalink / raw)
  To: libre-riscv-dev
In-Reply-To: <CAPweEDy0rN2kz=C-5jbZeVPVxVHCZ8OQWRneO26viTVcSJF03g@mail.gmail.com>

If no one else is working on the LDST unit I will start working there.
First I'll copy the skeleton files from another function unit, then I'll
try to implement the things described in the video L0 cache/buffer walkthrough [1].

[1] https://www.youtube.com/watch?v=6Yiyw4abJpE

On Sat, 23 May 2020 15:29:56 +0100
Luke Kenneth Casson Leighton <lkcl@lkcl.net> wrote:

> yesterday / night / morning:
> 
> * converted pipeline input and output stages to "common code"
> (soc.fu/*common*.py)
> * removed some unnecessary inputs and outputs as a result, from the pipelines.
> * created some more subset classes of Decode2ExecuteType
> * helped Jean-Paul understand the FU-FU Matrices for coriolis2
> * added ISEL to the Condition Register pipeline (including adding RB)
> * implemented CR_ISEL unit test and formal proof (michael, you
> duplicated the unit test)
> * began to make MultiCompUnit understand "regspecs"
> * documented where in MultiCompUnit that Cesar needs to put "zero_a"
> support (and how to make it optional)
> * added Michael's ssh key to the talos1 server
> * gave it a name (talos1.libre-soc.org - good idea, Jacob)
> 
> today:
> 
> * create a page describing "regspecs"
> * continue helping Cesar
> * look out for incoming questions and commits
> * watch at least 3 episodes from Season 4 of Stargate SG-1
> * see if there's any recent Rick and Morty or Fast n Loud.
> 
> l.
> 
> _______________________________________________
> libre-riscv-dev mailing list
> libre-riscv-dev@lists.libre-riscv.org
> http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev


-- 
Tobias Platen <libre-soc[at]platen-software[dot]de>

_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev

^ permalink raw reply

* [libre-riscv-dev] daily kan-ban update 23may2020
From: Luke Kenneth Casson Leighton @ 2020-05-23 14:29 UTC (permalink / raw)
  To: Libre-RISCV General Development

yesterday / night / morning:

* converted pipeline input and output stages to "common code"
(soc.fu/*common*.py)
* removed some unnecessary inputs and outputs as a result, from the pipelines.
* created some more subset classes of Decode2ExecuteType
* helped Jean-Paul understand the FU-FU Matrices for coriolis2
* added ISEL to the Condition Register pipeline (including adding RB)
* implemented CR_ISEL unit test and formal proof (michael, you
duplicated the unit test)
* began to make MultiCompUnit understand "regspecs"
* documented where in MultiCompUnit that Cesar needs to put "zero_a"
support (and how to make it optional)
* added Michael's ssh key to the talos1 server
* gave it a name (talos1.libre-soc.org - good idea, Jacob)

today:

* create a page describing "regspecs"
* continue helping Cesar
* look out for incoming questions and commits
* watch at least 3 episodes from Season 4 of Stargate SG-1
* see if there's any recent Rick and Morty or Fast n Loud.

l.

_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev

^ permalink raw reply

* Re: [libre-riscv-dev] Priority Encoder for Cesar
From: Luke Kenneth Casson Leighton @ 2020-05-23 14:21 UTC (permalink / raw)
  To: Libre-RISCV General Development
In-Reply-To: <48c27a2c-e121-01c0-31e7-85ce29ed9333@gmail.com>

On Sat, May 23, 2020 at 3:13 PM Cesar Strauss <cestrauss@gmail.com> wrote:
>
> Hi Luke,
>
> My assignments are piling up. I think I need a Priority Encoder...

:)

in order:

* (5+6) - this should not take more than 5 minutes to do the code
(commit it straight away, then we can review it together)

   a preliminary "proof-of-concept" unit test on the other hand might
take ... mm... 20 minutes?

* (4) - review, but consider it "done".

* (2) - L0CacheBuffer with two ports (not two CompLDSTs... yet) - is a
high priority unit test.

* (3) i can handle that... but only when (2) is done

* (1) after L0CacheBuffer unit test is done i expect this to show us
LDSTCompUnit is ok, but yes, needs more testing... but *after* (2).


> I have a wiki page now. I suppose I should update it with the above
> list, right?

yes, it is what it is for.

> I don't mind about the volume, but I would like to know on what to work
> first.

(5+6) high priority to get you "started".

(2) is the next one.

> I think I will start with 4 (review of code-morph), since it's the
> shortest, OK?

and is not involving code-development.  yes, review, but consider it "done".

_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev

^ permalink raw reply

* [libre-riscv-dev] Priority Encoder for Cesar
From: Cesar Strauss @ 2020-05-23 14:12 UTC (permalink / raw)
  To: libre-riscv-dev

Hi Luke,

My assignments are piling up. I think I need a Priority Encoder...

My tasks, in order of arrival:

1) Fix a bug in the LDSTCompUnit
https://bugs.libre-soc.org/show_bug.cgi?id=318
Status: Luke thinks he fixed it, but needs a review and improving the
unit tests. See: https://bugs.libre-soc.org/show_bug.cgi?id=318#c7

2) Test dual ports (two CompLDSTs with two ports, 4-4 as well) which
write to the same memory
https://bugs.libre-soc.org/show_bug.cgi?id=318#c11
Status: not started

3) Luke tried two LDs in the score6600 code - they failed.
https://bugs.libre-soc.org/show_bug.cgi?id=318#c17
Status: not started, need to check the [prototype] L0CacheBuffer

4) Code-morph LDSTCompUnit to use RecordObject structure, like CompUnitALU
https://bugs.libre-soc.org/show_bug.cgi?id=318#c18
Status: Need a review of Luke's implementation, compared to mine.

5) ALU CompUnit needs to recognise that RA (src1) can be zero
https://bugs.libre-soc.org/show_bug.cgi?id=336
Status: not started

6) Something about the above (5), being optional.
https://bugs.libre-soc.org/show_bug.cgi?id=336#c5
Status: not started

I have a wiki page now. I suppose I should update it with the above
list, right?

I don't mind about the volume, but I would like to know on what to work
first.

I think I will start with 4 (review of code-morph), since it's the
shortest, OK?

Regards,

Cesar

_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev

^ permalink raw reply

* [libre-riscv-dev] [Bug 332] Formal correctness proof needed for CR pipeline
From: bugzilla-daemon @ 2020-05-23 13:40 UTC (permalink / raw)
  To: libre-riscv-dev
In-Reply-To: <bug-332-13@https.bugs.libre-soc.org/>

https://bugs.libre-soc.org/show_bug.cgi?id=332

--- Comment #6 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
(In reply to Michael Nolan from comment #5)
> (In reply to Luke Kenneth Casson Leighton from comment #4)
> > michael could you check this when you have a moment?
> > 
> > commit 27440addbb83c3c13949246b495b8c474b1c2c90 (HEAD -> master,
> > origin/master)
> > Author: Luke Kenneth Casson Leighton <lkcl@lkcl.net>
> > Date:   Sat May 23 11:58:43 2020 +0100
> > 
> >     add CR_ISEL formal proof to CR pipeline
> 
> I changed it so the proof of isel uses the full CR register.

ok, cool.  we do now have two CR_ISEL case statements, and now two test_isel().

-- 
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev

^ permalink raw reply

* [libre-riscv-dev] [Bug 332] Formal correctness proof needed for CR pipeline
From: bugzilla-daemon @ 2020-05-23 13:28 UTC (permalink / raw)
  To: libre-riscv-dev
In-Reply-To: <bug-332-13@https.bugs.libre-soc.org/>

https://bugs.libre-soc.org/show_bug.cgi?id=332

--- Comment #5 from Michael Nolan <mtnolan2640@gmail.com> ---
(In reply to Luke Kenneth Casson Leighton from comment #4)
> michael could you check this when you have a moment?
> 
> commit 27440addbb83c3c13949246b495b8c474b1c2c90 (HEAD -> master,
> origin/master)
> Author: Luke Kenneth Casson Leighton <lkcl@lkcl.net>
> Date:   Sat May 23 11:58:43 2020 +0100
> 
>     add CR_ISEL formal proof to CR pipeline

I changed it so the proof of isel uses the full CR register. Part of the reason
that proof is written the way it is, is I'd already proven that the CR
operations worked when we were using the old interface where it would always
take in the full CR. I wanted to prove that the new way was exactly equivalent
to the old way, so I kept the old proof machinery that used the full CR
register, and added on a way to translate the new interface to that. I think
it's useful to keep doing it that way, so I modified your isel proof to do the
same thing.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev

^ permalink raw reply

page:   latest

Libre RISC-V Mailing list Archives (accessible via public-inbox)

Archives are clonable: git clone --mirror https://inbox.libre-soc.org/libre-riscv-dev

Example config snippet for mirrors


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git